Microsoft Discloses CVE-2026-45497, a Command-Injection RCE in 365 Copilot Already Fixed Server-Side With No Customer Action
Microsoft rated the Copilot command-injection flaw Critical, with a 7.7 CVSS base score. It was already mitigated in the cloud and not exploited.
Overview
Microsoft disclosed a remote code execution vulnerability in Microsoft 365 Copilot, tracked as CVE-2026-45497, in an advisory released on June 4, 2026, according to the National Vulnerability Database. According to Microsoft Security Response Center, the advisory describes the flaw as “Improper neutralization of special elements used in a command (‘command injection’) in Microsoft Copilot” that “allows an authorized attacker to execute code over a network.” Microsoft, the issuing CVE Numbering Authority, says the vulnerability had already been fully mitigated in its cloud service before the disclosure, and that customers need to take no action.
What We Know
The vulnerability affects Microsoft 365 Copilot, with an impact classified as remote code execution, according to Microsoft Security Response Center. Microsoft, the CNA, rates the issue Critical. The CVSS 3.1 base score is 7.7, with the vector string CVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:C/C:H/I:L/A:L, as listed by both Microsoft Security Response Center and the National Vulnerability Database. The same base score and vector are mirrored by Tenable. The flaw is categorized as CWE-77, “Improper Neutralization of Special Elements used in a Command (‘Command Injection’),” according to the National Vulnerability Database.
The vector indicates the attack is reachable over a network (AV:N) and requires no user interaction (UI:N), but carries high attack complexity (AC:H) and requires low privileges (PR:L), per the vector string published by the National Vulnerability Database. The scope metric is marked Changed (S:C), which in CVSS denotes that the impacted component can differ from the vulnerable one.
Microsoft states that the vulnerability was not exploited and had not been publicly disclosed at the time of the advisory, according to Microsoft Security Response Center. Because the fix was applied to Microsoft’s cloud infrastructure rather than shipped as an update, the advisory lists no patch package or KB article to install. In its FAQ, Microsoft explains: “This vulnerability has already been fully mitigated by Microsoft. There is no action for users of this service to take. The purpose of this CVE is to provide further transparency,” per the Microsoft Security Response Center advisory.
The National Vulnerability Database lists the record as published on June 4, 2026 and last modified on June 5, 2026, with Microsoft Corporation as the source of the CVSS assessment.
What We Don’t Know
The advisory does not name a researcher; Microsoft lists no acknowledgments for the finding, according to Microsoft Security Response Center. It also does not detail the specific command path, the privileged context required, or the technical steps an attacker would have needed to trigger the flaw. The National Vulnerability Database notes its own record is still being enriched and that an independent NVD assessment had not yet been provided beyond the CNA-supplied metrics.
Analysis
The disclosure reflects a pattern Microsoft has adopted for vulnerabilities in its cloud-hosted services: assigning a CVE for transparency even when there is no customer-facing update, because the fix is deployed entirely on the vendor’s side. As stated in the advisory’s FAQ, the CVE exists “to provide further transparency” rather than to prompt remediation, per Microsoft Security Response Center. One nuance worth flagging for readers tracking severity: Microsoft labels the issue Critical, while the underlying 7.7 base score falls within the High band used by the National Vulnerability Database and is shown as High severity by Tenable.